The intertidal zone is a dynamic environment, which results in the relatively frequent change of high and line water marks. It is important to be able to monitor these changes, as intertidal stability or instability can have a considerable influence on the rate and extent of coastal erosion and flooding, which are both expected to worsen with climate change.
Bespoke tidal surveys to capture the full extent of the intertidal is often expensive and logistically difficult, especially in areas with large tidal ranges. However, an approach developed within the Scottish Dynamic Coast project using Sentinel 2 data offers insight into the intertidal zone, quickly and easily.
Using a time-series of Sentinel 2 images for an area of interest, with each image capturing a slighlty different tidal position. We can then identify the water in each image (using the Normalised Difference Water Index (NDWI)). It is then possible to create an image that represents the frequency of water occurrence across the intertidal zone. This can be used to inform assessments of intertidal geomorphology, as shown in the image below:
Areas that are always water (i.e. areas below Low Water) will always be covered by water and have a water occurrence frequency of 100% (i.e. in all of the images in the image collection, water was always identified in that location). Areas that are sometimes covered by water will represent the intertidal zone. An example of the Coast X-Ray output for St. Andrews, Fife, Scotland, is shown below:
A fully interactive map based version can be see here.
This GitHub respository supports the journal publication, and includes the code and instructions to create a Coast X-Ray water occurrence output for your area of interest.
Coast X-Ray utilises Google Earth Engine (GEE) to produce the outputs. This is a very powerful tool and if you are a new user of GEE, it is highly recommended that you take a look at the introduction material and to become familiar with the GEE code editor and GEE terminology before proceeding with the Coast X-Ray code.
GEE uses JavaScript within the online code editor, however, added functionality (mainly associated with the exporting of assets) can be achieved by using Python via the Python API. Coast X-Ray uses the Python API, however, there is also some code which runs within R.
It can be sometimes difficult, especially for new users, to understand what the code does and for what purpose. Below is a succinct explanation of the processes that are carried out by the code. Note, that for the code to function correctly these steps need to be performed in sequence.
Step 1. Generate a grid grid/analysisGrid.ipynb
In order to make processing within GEE more efficient areas of interest are broken up into small areas using a grid. Coast X-Ray uses a Discrete Global Grid (DDG) to split the globe into small areas. Coast X-Ray makes use of the ISEA3H: Icosahedral Snyder Equal Area Aperture 3 Hexagonal Grid which can be accessed via the R package dggridR. A hexagonal grid was chosen as this ensures that (almost) all cells are of an equal area, regardless where you are on the globe. The ISEA3H grid can be created at different resolutions (i.e. cell sizes), with an example of given below of the 5th resolution:
The code within globalGrid/notebooks/globalGrid.ipynb generates a global grid of hexagons, at the users desired resolution, which is then saved locally. The code runs in a Jupyter Notebook, but note that it is within an 'R' kernel. A grid resolution of 12 is recommended as a good balance between size and efficiency.
Step 2 - Generate an intertidal grid intertidalGrid/intertidalGrid.ipynb
The global grid created in Step 1 is large and needs to be filtered down to an area of interest. However, the code within grid/intertidalGrid.ipynb not only filters the grid cells to the area of interest, but also to the cells that contain intertidal areas. This means that a whole country can be used as an area of interest, however only the intertidal/coastal cells will be selected for use in the subsequent analysis.
To identify the intertidal/coastal cells three supporting datasets are used: data on the intertidal area Murray et al. 2019, bathymetry data GEBCO, and the OpenStreetMap coastline.
Step 3 - Generate water occurrence analysis waterOccurrence/waterOccurrence.ipynb
The intertidal grid that was generated in the Step 2 will now be used to produce a water occurrence analysis. Each of the cells within the intertidal grid are processed in turn by the script in waterOccurrence/waterOccurrence.ipynb, producing a separate image for each cell. The images are added to GEE within an image collection, which can support filtering of the images and mosaicking.
For each cell within the intertidal grid, the water occurrence image is produced by:
- clipping the Sentinel 2 image collection to the grid cell
- filtering out images that do not cover all of the grid cell area (due to cloud cover or being at the edge of an image)
- calculating the Normalised Difference Water Index (NDWI) for each image
- identifying the water in each image using a fixed threshold approach
- amalgamating the time series of images to produce a single image by calculating the water occurrence percentage
- exporting the image to an image collection on GEE
- exporting image metadata to a feature dataset
Further Analysis
Using the image metadata output from step 3, if you have available to you via a tidal model or a tide gauge, a tide stage can be assigned to each image in a grid cell using the data/time of image aquisition.
Coast X-Ray was developed using the [NOC POLPRED] (https://noc-innovations.co.uk/software/offshore) model API under license. Therefore, code to allocate a tide stage to each image and the further processing is not provided here. However, the freely accessible FES2014 tidal model has been used by others and a version of Coast X-Ray is in development.
To use the code within Coast X-Ray the simplest approach is to create an environment in which to run the code. The environment mimics the Python environment that the code was developed upon, therefore allowing the code to run without errors. This environment is created and managed using Anaconda. The following instructions will guide you through the steps to install Anaconda and use Coast X-Ray.
GEE also offer a guide on installing GEE in an Anaconda environment and information on the GEE Python API.
Step 1: Download Coast X-Ray: Clone or download the Coast X-Ray repository from GitHub using the button 'Clone or Download' above and save it somewhere locally on your computer, e.g. C:\CoastXRay.
Step 2. Anaconda: Install Anaconda. Open the Anaconda prompt (PC) or on a Mac or Linux system open a terminal window. Use the cd command to change the directory, and navigate to go the folder where you have downloaded the Coast X-Ray repository (see here for information on how to use the command prompt).
Step 3: Create a new environment: Create a new environment named coastxray with all the required packages:
conda env create -f environment.yml -n coastxray
This environment should ensure that all the required packages are installed. This environment is called coastxray. The next step is to active the environment in the command prompt with:
conda activate coastxray
The terminal command line prompt should now have changed from (base) to (coastxray).
To use GEE you need to create an account. Go to https://earthengine.google.com and sign up. Once you have an account, in the Anaconda prompt run:
earthengine authenticate
This will open a web browser and ask you to login into your GEE account. An authorisation code will be given, which you copy and paste back in the Anaconda prompt.
This completes the installation.
The scripts are designed to run in sequence analysisGrid -> intertidalGrid -> waterOccurrence. They are also designed to run in Jupyter Notebooks. This is a "web-based interactive development environment" and allows for comments and descriptions to be added to the code to support/aid understanding.
Ensure that you have navigated to the CoastXRay directory within the Anaconda prompt, e.g. C:\CoastXRay, and you have activated the Coast X-Ray environment, then type:
jupyter notebook
Activating the environment and starting jupyter notebooks should look like this in your Anaconda Prompt:
This will then open a web browser, and you should be able to see the files within the C:\CoastXRay directory. The files that end with .ipynb contain the scripts. For more information on how to use Jupyter Notebooks, see this video.
You should use the scripts in the following order:
- grid/analysisGrid.ipynb
- intertidalGrid/intertidalGrid.ipynb
- waterOccurrence/waterOccurrence.ipynb
If you are having a problem, please create a new Issue. This keeps all problems and solutions in the same place and acts as a resource for other users to address the same/similar problems.
Coast X-Ray was developed by Dr. James M. Fitton, Dr. Alistair Rennie, Dr. Jim Hansom, Freya Muir, and Dr. Martin Hurst as part of the Scottish Dynamic Coast project.
The authors would like to thank Gennadii Donchyts and Rodrigo E. Principe for developing various elements of open-access code that was included within Coast X-Ray code, and Kilian Vos for open-access to CoastSat whose GitHub was a template for the development of the Coast X-Ray GitHub.
